Study proposes joint output-attribution regularization for fairness in medical image classification

A new arXiv paper argues output-level fairness metrics do not ensure demographic groups rely on similar predictive contributions, proposing joint output-attribution regularization for medical image classifiers.

Read the full story at arXiv cs.LG ↗

Timeline · 1 report

1. 2026-11-11 05:00 · arXiv cs.LG
   Explanation-Aware Fairness Beyond Output Parity: Joint Output-Attribution Regularization for Medical Image Classification